/*
* Recursive method that searches over directory structure
* to find the files. Then reads the contents of each file, and
* calls addToCollectionFreq() for each word.
* Calls appropriate FileMatch method for whether
* the search is being done with database or without.
* if true is returned, writeToOutput() is called.
* @params string dir, Boolean useDataBase, string[] searchTerms,
* @params List<HashSet<string>> newSearch
*/
public void SearchDir(string dir, Boolean useDataBase, string[] searchTerms, List <HashSet <string> > newSearch)
{
try //handles TextReader and StreamReader errors and non existing-file paths.
{
foreach (string file in Directory.GetFiles(dir)) //iterates over files in the current directory
{
String line; //variable to hold output from StreamReader
TextReader tr = new StreamReader(file); //reads the file
while ((line = tr.ReadLine()) != null) //fires if line is not null
{
String aLine = Regex.Replace(line, @"[^\w\s]", ""); //removes punctuation, and replaces it with "".
words = aLine.Split(' '); //sets an array to the line split at spaces.
for (int i = 0; i < words.Length; i++)
{
if (words[i] != "") //fires if word[i] is not "".
{
words[i] = words[i].ToLower(); //sets the words to lowercase
String wordi = words[i]; //sets the String variable wordi to words[i]
TermFreq.AddToCollectionFreq(wordi); //adds the word to collectionWordFreq Hashtable.
}
}
}
if (useDataBase) //fires if include synonym checked
{
if (Search.FileMatchTermDb(words, newSearch)) //fires if file contains a search term or its synonym
{
writeToOutput(file); //Writes filepath to output ArrayList
}
}
else//fires if include synonym unchecked
{
if (Search.FileMatchTerm(words, searchTerms))//fires if file contains a search term
{
writeToOutput(file);//Writes filepath to output ArrayList
}
}
}
foreach (string dirs in Directory.GetDirectories(dir)) //iterates through dirctories in directory structure.
{
SearchDir(dirs, useDataBase, searchTerms, newSearch); //calls its self, so files in next directory can be searched.
}
}
catch (Exception err)//catches TextReader and StreamReader errors
{
MessageBox.Show("" + err);
}
}
/*
* Populates form with filepaths, most common word,
* query term frequency and files found.
* Fires resets for collectionFreq and output,
* so they are new objects for next search.
* @param Boolean useDatabase.
*/
private void fileToForm(Boolean useDatabase)
{
if (fileUtil.getOutput.Count == 0)//handles no files being found.
{
FilePathOutput.Items.Add("No files found");
}
foreach (string s in fileUtil.getOutput) //iterates through output, using it's getter.
{
FilePathOutput.Items.Add(s); //adds each string to list box.
}
Found.Text = (fileUtil.getOutput.Count).ToString(); //sets Found.Text to length of output, using it's getter.
commonWord.Text = TermFreq.CollectionFreq(); //sets commonWord.Text to return variable from CollectionFreq()
QueryTermFreq.Text = TermFreq.QueryTermFreq(searchTerms); //sets QueryTermFreq to variable returned from QueryTermFreq()
fileUtil.ResetOutput(); //sets ArrayList output to new ArrayList.
TermFreq.ResetFreq(); //sets Hashtable collectionWordsFreq to new HashTable.
}
/*
* Sets the arrays for methods.
* @param string line (the contents of file)
* @param string file (filepath)
* @return string[]
*/
static private string[] SetArray(string line, string file)
{
string[] words; //sets new array
String trimLine = line.Trim(); //trims whitespace, /r and /n from start and end of line.
String aLine = Regex.Replace(trimLine, @"[^\w\s _]", ""); //removes punctuation, and replaces it with "".
words = aLine.Split(' '); //sets an array to the line split at spaces.
for (int i = 0; i < words.Length; i++)
{
if (words[i] != "" || words[i].Length < 50) //fires if word[i] is not "".
{
words[i] = words[i].ToLower(); //sets the words to lowercase
String wordi = words[i]; //sets the String variable wordi to words[i]
TermFreq.AddToCollectionFreq(wordi); //adds the word to collectionWordFreq Hashtable.
}
}
return(words);//returns words array to method that called SetArray()
}
/*
* Populates form with filepaths, most common word,
* query term frequency and files found.
* Fires resets for collectionFreq and output,
* so they are new objects for next search.
* @param Boolean useDatabase.
*/
private void FileToForm()
{
Dictionary <string, int> output = index.WriteToOutput();
if (output.Count == 0)//handles no files being found.
{
FilePathOutput.Items.Add("No files found");
}
foreach (string s in output.Keys) //iterates through output, using it's getter.
{
FilePathOutput.Items.Add(s); //adds each string to list box.
}
Found.Text = (output.Count).ToString(); //sets Found.Text to length of output, using it's getter.
commonWord.Text = TermFreq.CollectionFreq(); //sets commonWord.Text to return variable from CollectionFreq()
QueryTermFreq.Text = TermFreq.QueryTermFreq(searchTerms); //sets QueryTermFreq to variable returned from QueryTermFreq()
ICollection key = InvertedIndex.invertedIndex.Keys; //remove before submission, for debugging.
}
/*
* Resets data structures for rebuild
*/
public void ResetInvertedIndex()
{
invertedIndex.Clear(); // clears contents of invertedIndex
TermFreq.ResetFreq(); //resets collectionWordsFreq
}